ITU and Psychiatry: A Single-Axiom View of K_brain Failures, Predictive Coding, Drug Mechanisms, and the 2026-2050 Roadmap

We apply the Information-Theoretic Unification (ITU) framework (Terada 2026, DOI 10.5281/zenodo.20109210) to psychiatric biology. Psychiatric disorders are reframed as K_brain-component failures in the predictive-coding machinery; Friston's Free Energy Principle (FEP) is shown to be the brain-specific incarnation of the ITU axiom dS = d<K>. This is Tier 1 paper #7, completing the ITU Medicine Triangle: Cancer (#5, acute) + Aging (#6, chronic) + Psychiatry (#7, brain-circuit failures). Phase 67: ITU foundation. Eight major psychiatric disorders (schizophrenia, depression, anxiety, PTSD, ASD, ADHD, OCD, bipolar) are mapped onto nine K-component axes (perception, precision, reward, threat, attention, action, social, mood, self-model). Global disease burden is ~136.8 million DALYs (GBD 2021, ~7% of all DALYs; $5T global economic burden). Treatment success correlates inversely with K-identification difficulty (ADHD 70% response with clear K_attention target; ASD 30% with rigid K_social). Phase 68: Schizophrenia as K_precision failure - top-down prior dominates bottom-up sensory evidence, producing hallucinations and delusions. A Bayesian belief model with precision ratio 0.3 (vs healthy 1.0) produces belief lock-in. Treatment response by symptom group: positive 65-78%, negative 15-45%, cognitive 10-35%. The Howes-Kapur dopamine asymmetry (mesolimbic excess + mesocortical deficit) is reproduced. Six antipsychotics compared; Clozapine rescues ~55% of TRS (30% of all patients). KarXT (2024 FDA approval) opens a non-D2 K_precision restoration path. Phase 69: Depression as K_reward collapse - positive prediction errors not registered (anhedonia). After 100 events, K_reward belief settles at 0.28 (severe depression) vs 0.49 (healthy). Anxiety/PTSD = K_threat over-precision; false-positive threat detections explode (97/200 anxiety, 153/200 PTSD vs 15/200 healthy). Antidepressant kinetics compared: SSRI 4-6 weeks, ketamine hours (88% at day 1, NMDA), psilocybin one-session durable (70%+ at week 4, 5-HT2A), ECT cumulative 75% at 4 weeks. STAR*D-like cascade: 37% (Step 1) to 70% (Step 5 + rapid-acting/ECT) remission; TRD at 30%. Phase 70: ASD (K_social rigid over-precision, sensory hyperresponse) and ADHD (K_attention filter failure with SNR 1.82 vs healthy 12.21; stimulants restore SNR to 5.60). Digital phenotyping (Apple Watch, smartphone passive sensing) enables continuous K-monitoring. Brain stimulation (TMS, DBS, focused ultrasound, vagus) and psychedelic-assisted therapy (MDMA-AT, psilocybin, ketamine) complete the 4-axis treatment paradigm. 2026-2050 roadmap predicts psilocybin FDA approval 2027, MDMA-AT re-approval 2026, FUS for depression 2028, K-monitoring standard 2030, and DSM-6 K-component-based diagnosis 2040. Ten falsifiable predictions issued. Central thesis: psychiatric disorders are K_brain-component failures requiring multi-axis (drugs + therapy + digital + brain stimulation) restoration - paralleling Cancer (Tier 1 #5) and Aging (Tier 1 #6) multi-K therapy patterns. Honest framing: this is a Pass-1 interpretive paper that reframes computational psychiatry (Friston FEP, Bayesian Brain, dopamine hypothesis, STAR*D, treatment cascades) in ITU language. Pass-2 work would derive an ITU-specific EEG/MRI biomarker. This completes the ITU Medicine Triangle, joining the Engineering Rectangle (Quantum Computing 10.5281/zenodo.20139391 + Machine Consciousness 10.5281/zenodo.20150501 + Cryptography 10.5281/zenodo.20151059 + Semiconductors 10.5281/zenodo.20174036) plus Cancer (10.5281/zenodo.20174318) and Aging (10.5281/zenodo.20175663) to form ITU's first complete polytope structure. Includes 4 theory documents, 4 Python numerical experiments, 4 figures, 4 JSON summaries. Total runtime ~35 seconds.